English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Discriminative analysis in neuroimaging by means of deep/machine learningtechniques is usually tested with validation techniques, whereas the associatedstatistical significance remains largely under-developed due to theircomputational complexity. In this work, a non-parametric framework is proposedthat estimates the statistical significance of classifications using deeplearning architectures. In particular, a combination of autoencoders (AE) andsupport vector machines (SVM) is applied to: (i) a one-condition, within-groupdesigns often of normal controls (NC) and; (ii) a two-condition, between-groupdesigns which contrast, for example, Alzheimer's disease (AD) patients with NC(the extension to multi-class analyses is also included). A random-effectsinference based on a label permutation test is proposed in both studies usingcross-validation (CV) and resubstitution with upper bound correction (RUB) asvalidation methods. This allows both false positives and classifier overfittingto be detected as well as estimating the statistical power of the test. Severalexperiments were carried out using the Alzheimer's Disease NeuroimagingInitiative (ADNI) dataset, the Dominantly Inherited Alzheimer Network (DIAN)dataset, and a MCI prediction dataset. We found in the permutation test that CVand RUB methods offer a false positive rate close to the significance level andan acceptable statistical power (although lower using cross-validation). Alarge separation between training and test accuracies using CV was observed,especially in one-condition designs. This implies a low generalization abilityas the model fitted in training is not informative with respect to the testset. We propose as solution by applying RUB, whereby similar results areobtained to those of the CV test set, but considering the whole set and with alower computational cost per iteration.

Deep Learning in current Neuroimaging: a multivariate approach with power and type I error control but arguable generalization ability